This invention relates generally to gas analyzers for detecting one or more component gas concentrations in a sample and, more particularly, to non-dispersive infrared gas analyzers.
Non-dispersive infrared gas analyzers have long been known. Such analyzers direct a source of infrared radiation along an optical path in a preselected spectral band having at least one absorption line of the component gas to be detected. A detector, positioned in the optical path, detects radiation in the preselected spectral band and produces a detector output. A sample chamber is positioned in the optical path between the source and the detector in order to contain a quantity of the sample gas, which includes the component gas to be detected. Such infrared gas analyzers have suffered from numerous problems. Their response time, being on the order of one to three seconds, is much too slow for many applications. For example, it is not practical to accurately track a patient's respiration carbon dioxide level with an instrument having a response time of between one and three seconds. Furthermore, known systems have been plagued with temperature and age drift problems, which has resulted in costly and complicated compensation schemes. Such compensation schemes further limit the adaptability of the instrument because they increase the size and reduce the reliability of the instrument.
Sources for infrared gas analyzers radiate electromagnetic energy in the infrared band by applying a current to a resistance element, such as a tungsten alloy filament In order to increase the emissivity of the source, it is known to run the source at ever-increasing temperatures. One difficulty with such an approach is that it increases the oxidation of the source element. Such oxidation not only reduces the efficiency of the source, but results in a significant amount of drift in the source emissivity over time. This instability must be compensated for in order to avoid a resulting inaccuracy in analyzer readings. One approach at solving this problem has been to coat the element with a ceramic jacket in order to avoid oxidation of the elements. Although this approach may reduce oxidation of the element per se, various oxides are still formed and deposited on the ceramic layer resulting in undesirable variation in the source emissivity.
In order to overcome the degrading effects of source oxidation on analyzer accuracy, the prior art has attempted to sense the level of source emissivity by placing, for example, temperature sensors in a position to monitor the temperature of the source. Because source emissivity is proportional to temperature, the output of the temperature sensor may be used in a feedback loop to regulate the current applied to the source and, in theory, maintain a constant level of emissivity of the source. The principle behind this approach is that, if source emissivity is regulated to a constant level, system drift will then be kept to a minimum. In practice, this has not worked satisfactorily. The extensive amount of beat generated by the source introduces second and third order errors in the detector circuitry, as well as in the system's optics. These errors have proven to be significant in degrading system performance. An alternative approach has been to apply the output of the source temperature sensor to regulate the detector circuitry. This has proven to be equally futile for the same reasons. The heating of the system components by the source has introduced errors that may not be readily compensated for.
Another approach at applying compensation to a gas analyzer system has been to divide the radiation path either spatially into two beams, a reference beam and a sensing beam, or, temporally, into a sequence of intermittent transmissions between the source and detector separated by blank periods when no beam is being transmitted. The principle behind this approach is that the detection circuitry may provide an accurate reading by comparing the reference beam, or period, with the sensed beam, or period, in order to subtract the effect of source strength from the detected signals. One approach at temporally dividing the detected signal into intermittent pulses is by chopping the light source utilizing a mechanical vibrating device or rotating device. This approach has several difficulties. The requirement for a mechanical device not only adds to the bulk and complexity of the system, but also adds to the system's low reliability. Furthermore, the thermopile detectors that are commonly used in infrared gas analyzers respond very slowly to the large swing in the detected beam between the blank periods and the sensing periods. This contributes to a system response time on the order of one to three seconds for such systems.
Another approach to temporally dividing the beam into intermittent segments is by switching the source periodically between off and on conditions. Because source emissivity is a function of heat output, they are not readily switched between off and on conditions because the heat must dissipate between each interval. As a result, switched source gas analyzers have the same difficulty of slow system response time as do the mechanically chopped beam systems. Furthermore, the periodic switching of the source introduces temperature variations to the system components, which not only degrades the system's longevity but also make appropriate compensation impractical.
In order to calibrate an infrared gas analyzer, both zero, or offset, and span calibration procedures are performed. The zero calibration is in order to compensate for any offset in the amplifiers and is traditionally performed by filling the sample chamber with a gas, such as air or nitrogen, that does not absorb infrared radiation at the absorption line of the gas detected by the analyzer. With the non-absorbing gas filling the chamber, the offset of the output is measured and, either stored for compensation in the gas concentration reading, or is reduced to a zero value. The span calibration is traditionally performed by filling the sample chamber with a known concentration of the gas or gases to be detected. The output is adjusted to correspond with the known gas concentration. Such span calibration is time-consuming and utilizes expensive calibration gases which are often toxic. As a result, alternate span calibration techniques have been proposed but all suffer from some deficiencies. One known alternative technique is to partially occlude the source with a partial transmission filter, or the like, in order to simulate the introduction of a known concentration of the gas to be measured. However, the introduction of the occluding member requires a mechanical movement.
Another alternative proposal is to inject an electrical current into the output amplifier of the electronic control in order to simulate the effect of a calibration gas in the sample chamber. While such technique does not suffer the drawbacks described with respect to the other known span calibration techniques, it is only capable of calibrating the output amplifier. It is incapable of compensating for fouling of the surfaces of the sample chamber or degradation in source performance.
In addition to the zero and the span calibrations, both of which can be performed in the field, an additional procedure must be occasionally performed at the factory or a service center. This additional procedure is a relinearization of the instrument. This results from the degradation of the source and the fouling of the sample chamber, in addition to drift and the like in the electronics, due to component aging. Because the relationship between the detector output and the source input varies according to a non-linear relationship, any variations in the performance of the component require that the instrument be calibrated at multiple values of sample gas calibration. The necessity to periodically return the instrument to the factory or a service center results in a significant increase in the total operating expense of an infrared gas analyzer.